{"id":"W4416036180","doi":"10.18653/v1/2025.emnlp-main.830","title":"Not-Just-Scaling Laws: Towards a Better Understanding of the Downstream Impact of Language Model Design Decisions","year":2025,"lang":"","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; National Institute of Advanced Industrial Science and Technology; National Science Foundation","keywords":"Downstream (manufacturing); Natural language; Language model; Natural (archaeology); Empirical research; Natural language generation","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01154139,0.001112885,0.001455434,0.001245351,0.001406511,0.006591367,0.00253897,0.002379599,0.01643264],"category_scores_gemma":[0.1435435,0.001300484,0.001023744,0.00170684,0.003586123,0.02745303,0.003801038,0.007346522,0.002297662],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001823583,"about_ca_system_score_gemma":0.00228058,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004449144,"about_ca_topic_score_gemma":0.004215486,"domain_scores_codex":[0.9933968,0.003762301,0.0002845547,0.0009803607,0.001268896,0.0003071431],"domain_scores_gemma":[0.887044,0.08802734,0.004895578,0.01361065,0.005097416,0.001325085],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001757397,0.0002257106,0.005312537,0.0003133291,0.0001121893,0.0002071004,0.001189408,0.03751858,0.001640743,0.8461469,0.01999142,0.08716634],"study_design_scores_gemma":[0.00002311054,0.00002513194,0.0005635718,0.00005090498,0.00002536026,0.000040734,0.0001719977,0.1214705,0.0009213039,0.8721545,0.004526494,0.00002631979],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04662633,0.003447355,0.8818151,0.02143896,0.000615136,0.0001572816,0.0006601571,0.001632302,0.04360729],"genre_scores_gemma":[0.771352,0.002813686,0.2105631,0.003369346,0.0006478505,0.000411442,0.0007998268,0.002458559,0.007584126],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01643264,"threshold_uncertainty_score":0.06103736,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1244222083353944,"score_gpt":0.3400544565899107,"score_spread":0.2156322482545163,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}